BACKGROUND: Brainstem evoked response audiometry (BERA) is not widely used for hearing screening because it is considered less practical; however, it is often used for diagnostics. Since the founding of automated auditory brainstem response (AABR), it often uses because it is more practical, has a high sensitivity and specificity in early detection of hearing loss (HL) in high-risk infants.
 AIM: The objective of the study was to determine the differences results of AABR and BERA for HL detection in high-risk infants at neonatal intensive care unit (NICU).
 METHODS: The study was conducted from November 2014 to September 2015 with consecutive sampling. The subjects were high-risk infants treated in the NICU room of the Neonatology Division at Dr. Soetomo General Hospital Surabaya and examined using AABR or BERA to determine the existence of HL.
 RESULTS: BERA results obtained normal (negative) as many as 28 ears (73.68%) and not normal (positive) as many as 10 ears (26.32%). AABR results obtained pass (negative) as many as 23 ears (60.53%) and refer (positive) as many as 15 ears (39.47%). Detection of HL in high-risk infants in NICU with AABR obtained 40% of sensitivity and 60.71% of specificity, 26.67% of positive prediction (NPP), 73.91% of negative predictive value (NPN), 55.26% of accuracy, 39.29% of false positive error rate, and 60% of false negative error rate. The comparative test of Wilcoxon signed-rank between the results of AABR and BER obtained p = 0.236.
 CONCLUSIONS: There was no difference between AABR and BERA results for HL detection in high-risk infants at NICU.